The “R” in WAR

Matt Kemp’s defense has always made him closer to replacement level than he appears to be. (via Dirk Hansen)

In baseball’s system of advanced metrics, Wins Against Replacement has become the Grand Unified Theory for assessing players’ contributions to their teams. To call WAR a stat is to undersell it. It is a combination of stats used to answer the age old-question of sports discussions. Taking account of all of his measurable contributions, it attempts to quantify how valuable a player is to his team.

WAR has grown enough in popularity that even fans and writers outside the sabermetric community discuss it, as they did in the conversation over whether Miguel Cabrera or Mike Trout should be the American League’s MVP in 2012. Trout led in WAR and placed second in the voting, while Cabrera led in many of the traditional counting stats–including winning the Triple Crown–and finished first.

The debate brought WAR into the wider baseball world, but many still do not fully understand it. A tweet by ESPN baseball analyst Dan Szymborski‏ drove home that point in July, showing that many fans believe WAR measures a player’s performance against the league average:

Defining Replacement Level

Szymborski‏ is correct; we have done a good job at popularizing what WAR measures but a bad job at explaining what a player’s WAR is measured against. Rather that measuring against the average major league player, WAR measures against a replacement player. What does that mean? FanGraphs’ glossary entry on WAR explains it this way: “WAR offers an estimate to answer the question, ‘If this player got injured and their team had to replace them with a freely available minor leaguer or a AAAA player from their bench, how much value would the team be losing?’”

The difference is important. Szymborski‏’s critic thought WAR showed Matt Kemp to be having a perfectly league-average season in 2017 and a slightly above league-average season in the three years prior. What it actually shows: Kemp’s 2017 performance is equal to what you would expect from a random Triple-A player called up to take his place. Sabermetrician Tom Tango defined replacement level as “the talent level for which you would pay the minimum salary on the open market, or for which you can obtain at minimal cost in a trade.” For someone earning $21 million this year, as Kemp is, that’s not good.

WAR is calculated in different ways, with FanGraphs’ and Baseball-Reference’s dueling versions being known as fWAR and bWAR, respectively. Since 2013, however, the two sites have agreed on what replacement level means. For devoted members of the sabermetrics community, this was a momentous occasion, and the unified definition gives us a common starting point even though the different versions of WAR continue to be calculated differently.

A replacement-level player is not one who gives your team nothing. That guy would not be on a major league roster. Instead, he is a guy who, if your whole team was made up of similar guys, would get your team to 47.7 wins over a full season. Put another way, it is equal to 1,000 WAR per 2,430 major league games. There hasn’t been a team that bad since the 2003 Detroit Tigers, which finished 43-119, 47 games out of first place in the American League Central. That’s a bare minimum, and any positive WAR is a reflection of improvement over that level.

Even with that definition, “replacement player” conjures a fuzzy image: a man who is good enough to play major league baseball, but only if the team can’t get somebody better. If Kemp got hurt and the Atlanta Braves called up a middling outfielder from Triple-A Gwinnett to fill in for 10 days, a zero-WAR performance is what you would expect; not a star player or a prospect, just a guy who can field his position and hit a little bit. He’s not going to win many games for you, but he won’t cost you many, either.

That still does not give us a complete picture of a replacement player. The hypothetical outfielder who could replace Kemp has no face, no number, and no actual stats to discuss. But to get an idea of what a zero WAR guy might look like, we can turn to the stats of real-life players who meet the replacement player description.

Kemp’s Zero-WAR Season

Start with Kemp himself, whose 2017 season, through August, had produced a WAR equal to Bluto Blutarsky’s GPA: zero point zero (it has since dropped below replacement level). In 98 games through the end of August, Kemp’s batting average, on-base percentage, and slugging percentage were .291/.336/.473. Those numbers are respectable enough, and in line with his career averages, which is likely what led to the fan’s assertion that Kemp is having a fine season. A deeper dive into the numbers—something now possible with advanced metrics—shows the error of that belief.

Kemp has battled hamstring injuries all season, which are likely one cause of the greatest drop-off in performance, his defense. Going by the old-fashioned measurement of fielding percentage, Kemp’s defense (through last Friday) looks great, at .993. But WAR takes into account the vastly more effective Ultimate Zone Rating (UZR), which tells us not just how many balls the fielder mishandled, but his total defensive contribution, including his range at his position (for analyzing stats from the pre-sabermetric era, another stat called Total Zone (TZ) is used).

A Hardball Times Update
Goodbye for now.

Here, Kemp stumbles badly. His UZR so far in 2017 is -8.5. UZR is measured against a league average, so being that far into the red highlights how many runs Kemp’s defense has cost Atlanta this year. And he’s also playing the vastly less important position of left field, compared with other years when he played center field, a position higher up on the defensive spectrum. By playing the position lower on the spectrum, he gets less of a positional adjustment. Where center fielders are given a positional adjustment of +2.5 runs, left fielders are given a tweak of -7.5 runs. Just what is positional adjustment you ask? Here’s an excerpt on it from the FanGraphs library:

In general, we want to add runs for players who play tough positions and subtract runs for players who play easier positions to account for the fact that average at one does not equal average at the other in terms of total run prevention. For example, for a shortstop, the adjustment is +7.5 runs per full season. For a left fielder, it’s -7.5 runs per full season.”

So, Kemp is playing the easier position, which gives him less credit. But his baserunning further damages his team’s chances and drags down his WAR even more. By contrast, Braves center fielder Ender Inciarte has an above-average baserunning performance and a UZR of 1.5 so far this year, part of the reason his 2017 WAR is 2.7 despite an offensive performance lacking in power.

WAR shows us that Inciarte’s contributions on the field make him more valuable than a replacement player, while Kemp’s drag him down to replacement level (now sub-replacement level). Kemp’s fielding makes him equal not only to the imaginary replacement player, but also to players who have actually replaced him in left field last year, as Paul Swydan showed in this FanGraphs article from March.

Replacement Level Career?

Much of the disconnect over Kemp’s replacement-level performance is because the parts of his game that are failing—defense and baserunning—are not as obvious to the casual observer as hitting performance is. The situation is further complicated by Kemp being a historically good player. His career WAR of 22.9 is respectable, and his peak year with the Dodgers—8.3 WAR in 2011—was good enough for him to place second in Most Valuable Player Award voting that year. In the average fan’s mind, Matt Kemp is a good player, so the contention that his 2017 season is equal to an unknown Triple-A call-up is hard to process, even if it is demonstrably true.

So what would a player with zero lifetime WAR look like? What is the career arc of a man who embodies the sabermetrical definition of a replacement-level player? The player with the longest-lasting zero-WAR career was Oscar Stanage, a catcher who played more than 1,000 games with the Tigers between 1909 and 1925. His SABR biography describes him as “inconspicuous yet ever-present,” and notes that he “never batted higher than .264 in any individual season, finished his major league career with a .234 batting average and just eight home runs.”

Stanage’s best season by WAR was 1909 (1.1) and his worst was 1914 (-1.3). His offense was far below a league average that, in the deadball era, was low to begin with. So why did he play for so long? As a catcher, Stanage’s defensive abilities cancelled out his weakness with the bat, and his durability at the peak of his career made him a dependable presence behind the plate.

There are also, then and now, the intangibles WAR cannot measure. Stanage was a clubhouse leader who became a reliable veteran for his club, attributes that also balance out sub-par performance on the field. (There is often some disparity between fWAR and bWAR, as well, especially on the defensive side of things; Stanage’s fWAR is 0.0, but his bWAR is 6.0!)

That still leaves an unclear picture to the current fans of the game. MLB may refer to everything since 1901 as the “modern era,” but the game Stanage played is very different from the one we know today. Since Stanage’s day, Babe Ruth drastically altered the sport, and Jackie Robinson transformed it again. So who in present-day baseball is having a replacement-level career?

The most prominent active player with a zero lifetime WAR probably is shortstop Paul Janish. In nine seasons with the Cincinnati Reds, Atlanta Braves, and earlier this season, the Baltimore Orioles, Janish has flitted between the majors and Triple-A fairly often—something the definition of replacement level suggests he should do. Both fWAR and bWAR agree on Janish: They rate him at 0.0 and ‑0.1, respectively. His most recent call-up was in June from the Triple-A Norfolk Tides to replace the Orioles’ J.J. Hardy when Hardy suffered a fractured wrist. Janish was returned to Norfolk after 14 games of -0.5 WAR ball.

Like Stanage, Janish is a player whose defense at a valuable position makes up for his offensive shortcomings. As David Laurilla wrote for FanGraphs in 2015, Janish “is your classic good-glove, no-hit infielder.” His career UZR is 17.3, which means teams will plug him in at shortstop in a pinch. His .212/.280/.284 offense means he likely won’t stay there for long.

Not Bad, Just Not Great

Objective measures can sound harsh, and calling a man “replaceable” has the ring of cruelty, but a zero-WAR career is far from the worst thing one can say about a player. Bill Bergen, whom former FanGraphs writer Joe Pawlikowski justly called “the worst batter in baseball history” back in 2011, piled up a career WAR of -16.2 in 11 seasons between 1901 and 1911. That eye-popping deficit is almost impossible to imagine today. The worst player by WAR in recent history is Juan Castro, with a merely execrable -6.1. By those standards, Janish and Stanage aren’t that bad.

By another measure, Janish’s 0.0 performance is even better than Kemp’s current season. Janish’s 2017 contract with the Orioles is for the league minimum, $535,000. He served exactly the role the O’s intended for him when they signed him: a good Triple-A shortstop who could serve in the majors if necessary. Compare that with the production Kemp has provided for $21,750,000, and the Orioles look like sabermetric savants. Janish’s career equaling zero WAR gives us a good idea of what replacement level means, and it shows the bar has been set right about where it should be.

References & Resources

Kyle Sammin is a lawyer and writer from Pennsylvania. Read his other writing at his personal website, and follow him on Twitter @KyleSammin.
Newest Most Voted
Inline Feedbacks
View all comments
6 years ago

Willie Bloomquist and Dewayne Wise have been my last two faces of replacement level.

6 years ago
Reply to  Tangotiger

Bloomquist has nothing on Yuniesky Betancourt.

6 years ago
Reply to  Tangotiger

Don Kelly and Darnell McDonald for me

Jeff Long
6 years ago

Wanted to drop a note in here (a quote from the glossary actually) over at BP on how we define replacement players, because it:
1. Is different from how BRef and FG approach things
2. Seems unfortunate to omit the third common WAR(P) from the discussion

From the BP Glossary:
“Perhaps no sabermetric theory is more abstract than that of the replacement-level player. Essentially, replacement-level players are of a caliber so low that they are always available in the minor leagues because the players are well below major-league average. Prospectus’ definition of replacement level contends that a team full of such players would win a little over 50 games. This is a notable increase in replacement level from previous editions of Wins Above Replacement Player.”

The concept of replacement player was initially proposed by Keith Woolner while at BP when he rolled out Value Over Replacement Player (VORP) many years ago. Rob Neyer did an excellent profile on Woolner and VORP ( That I think is a good read for anyone looking to understand how the concept of replacement players came about and was popularized.

Dave Studemanmember
6 years ago
Reply to  Jeff Long

Just want to add a footnote to Jeff’s comment. The replacement player concept was, as far as I know, first proposed by Bill James in the 1980’s, and Bill doesn’t claim he thought of it himself. Someone may have already written about it before Bill and he might have just picked up on it. It’s been around a long time.

I think Colin Wyers or Patriot wrote about this at BP a while ago. I don’t remember which article.

6 years ago

Careful: a team of replacement-level players isn’t worth 1000 WAR over 2430 games. They’re worth 0 WAR! (As they are, by definition, replacement-level)

Michael Bacon
6 years ago

You write, “He’s not going to win many games for you, but he won’t cost you many, either.” As John McEnroe was fond of saying, “You cannot be serious!” If a team of replacement players would theoretically win only 47.7 (try explaining that extra 0.7 to someone not versed in Sabermetrics and watch the lost look develop on their face, and some “stat nerds” wonder why the general Baseball fan ridicules we “stat nerds”…) wins out of 162 games that means the team would finish with a record of 47.7 – 114.3, so how about rounding it off to “48”? This means that any time during the season a replacement player takes the field he is costing his team WINS. The more he plays the more he will “cost” the team. Your premise has therefore been refuted.

It has been shown from a lifetime devotion to Baseball that the majority of replacement players are adequate fielders but lack hitting ability. An example would be Rafael Belliard, who was used by the brillian manager of the early ’90’s Braves in conjunction with Jeff Blauser, who was better at the dish than in the field, at SS. Raffy was in The Show because of his glove, not his bat. Managers will sacrifice hitting ability for defense when it comes to the bench, as has been shown historically. A later Braves manager, Freddie, had a sign the players read before hitting the field which said, “DEFENSE WINS.” No team shows this better than the 1969 New York Mets, who allowed an astoundingly low .257 BABIP against. The ’69 Braves, who won the Western division (The braves being in the Western division shows just how stupid can be the pooh-bahs of MLB), were second in the NL that year with a BABIP against of .267. DEFENSE WINS! If you do not believe me, please check out the late ’60’s-early ’70’s Baltimore Orioles. Their DEFENSE made average pitchers look GREAT! Then look at how badly the Pittsburgh Pirates of ’69 were hurt by their team defense. Their team BABIP against was .295. Bob Veale, a great pitcher, was victumized to the tune of a BABIP against of .343! He was a lefty who naturally faced more right handed batters, and the Pirates SS-3B combo of Freddie Patek and Richie Hebner were atrocius in the field. You can look it up. DRA shows Patek costing his team 40, FORTY, runs that season. I believe that works out to 80, EIGHTY, more balls getting through for hits. Add Hebner’s 10 runs allowed, or 20 hits that would have been turned into outs by an average fielder, and the Pirates, who could have been in contention, became an also ran.

As for Matt Kemp, although true that he has had hammy problems this year, his defense last year, a relatively injury free season, was deplorable. This has made some of the Braves pitchers look bad, or at least much worse than they were in actuality.

6 years ago

“showing many fans believe WAR measures a player’s performance against the league average”

Really good article, but I’m not sure I agree with this. (Also, the tweet literally only shows that one person thinks that.)

I agree that WAR can be difficult to fully grasp, but IMO almost anybody with a cursory knowledge of WAR is at least aware of the difference between “replacement” and “average.”

I think it’s a pretty well-known rule of thumb that “average” position player production is about 2 WAR over the course of a season.

Paul G.member
6 years ago

One of the things I liked about Win Shares is it did not have the 0.0 issue. In Win Shares, 0 was terrible. Part of this was because it was not possible to score negative Win Shares, but even with modified systems with negative WS there were not many players below 0 and typically not more than the low single digits. WS had plenty of issues, but it got that right.

Any system where 0 is not really, really good or really, really bad is going to cause confusion. WAR compounds this by having a vague term for “replacement” which is more or less “bad but useful, I guess.” I do wonder if a WAZ (Wins Above Zero) is even possible.

6 years ago
Reply to  Paul G.

A 0 WAR player would get paid close to the league minimum in an open market.

6 years ago

If you need more convincing you can head over to And you will find that Matt Kemp is 59 out of 59 in ML OF with 450+ plate appearances this year. That is NOT average. Average would be positions 15-20; Avisail Garcia, Josh Reddick, Brett Gardner, Marwin Gonzalez, and JD Martinez. JD Martinez is a much better hitter to be considered average but his defense is well below average which hurts his overall WAR

6 years ago

bWAR is ambiguous, because it fails to differentiate BP and BR; generally better to use rWAR instead

Michael Bacon
6 years ago

To Paul G, in answer to, “I do wonder if a WAZ (Wins Above Zero) is even possible.”
Not only is it possible, but Baseball-Reference has a Wins Above Average, or WAA, on it’s website and I much prefer it to WAR. Although I understand WAR, it is still some kinda nebulous thing. With WAA one knows that a player with a score of +3 would make a 81-81 team a 84-78 team. I like the simplicity.